顾及属性相似性和空间异质性的滇中受污红土纳微过程模拟

Nano-Micro Process Simulation of Contaminated Laterite in Central Yunnan Considering Attribute Similarity and Spatial Heterogeneity

摘要: 空间复杂系统具有明显的空间结构异质性，其时空过程模拟中，属性相似性和空间异质性是过程模拟系统是否能正确表达过程中规律的两个重要影响因素。不同尺度不同逻辑域内的空间划分，是空间聚类或分类的重要制约因素，对属性相似性和空间异质性有关键影响。如何在确保特定空间场景基本格局的系统性与完整性合理表征的情况下，确保属性相似性的精度和空间异质性的准确度，以建立可靠模拟模型，表征时空过程的不同尺度或不同程度的规律，一直是跨域跨尺度复杂系统时空建模与模拟分析研究中亟待解决的一个重要科学问题。基于特殊红土纳微过程模拟，对不同尺度上的空间异质性进行全局空间自相关分析和局部空间自相关分析，同时采用核密度分析创建二维热力图对不同尺度上的空间异质性进行直观表达，然后提出采用跨域尺度不变特征进行多智能体系统(Multi-Agent System, MAS)建模，通过在酸碱不同逻辑域内探查尺度不变特征，然后查找跨域尺度不变特征，最后利用尺度不变特征进行MAS模拟，采取重构的特征图像与对应养护天数的原始特征图像进行有效性验证，从而确定了采用跨域尺度不变特征构建MAS模型的有效性。

Abstract: Spatially complex systems have obvious spatial structural heterogeneity. In the simulation of their spatiotemporal processes, attribute similarity and spatial heterogeneity are two important factors affecting whether the process simulation system can correctly express the laws of the process. Based on the simulation of special laterite nano-micro processes, the project conducts global spatial autocorrelation analysis and local spatial autocorrelation analysis on spatial heterogeneity at different scales. At the same time, it uses kernel density analysis to create a two-dimensional heat map to intuitively express spatial heterogeneity at different scales. Then, it is proposed to use cross-domain scale-invariant features for Multi-Agent System modeling to take into account attribute similarity and spatial heterogeneity, by exploring scale-invariant features in different logical domains of acid and alkali, and then finding cross-domain scale-invariant features, and finally using cross-domain scale-invariant features for MAS.